Bifurcation analysis for a SIRS model with a nonlinear incidence rate

Basic Information

• Paper ID: 2510.27292
• Title: Bifurcation analysis for a SIRS model with a nonlinear incidence rate
• Authors: Xiaoling Wang, Kuilin Wu (School of Mathematics and Statistics, Guizhou University)
• Classification: math.DS (Dynamical Systems)
• Publication Date: October 31, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.27292

Abstract

This paper conducts an in-depth investigation of the SIRS epidemic model with a general nonlinear incidence rate $f(I)S=\beta I(1+\upsilon I^{k-1})S$ ($k>0$). The authors systematically analyze the existence and stability of equilibrium points and apply local bifurcation theory to explore the rich dynamical behaviors of the model. The paper derives normal forms for different types of bifurcations, including Bogdanov-Takens bifurcation, nilpotent focus bifurcation, and Hopf bifurcation. By computing the first four focal values, the authors determine the codimension of Hopf bifurcation and reveal complex dynamical phenomena where the system may generate multiple limit cycles. Numerical simulations validate the correctness of the theoretical analysis.

Research Background and Motivation

Research Problem

Bifurcation phenomena in infectious disease dynamics refer to sudden changes in transmission dynamics or epidemic behavior caused by environmental, host, or pathogen factors. These bifurcations not only determine the speed and scope of transmission but may also alter viral pathogenicity, thereby affecting the effectiveness of public health interventions.

Problem Significance

1. Practical Importance: Accelerated globalization and ecological transitions have led to increasingly complex disease transmission patterns. Bifurcation analysis is crucial for predicting epidemic trends, formulating effective control strategies, and efficiently allocating public health resources.
2. Theoretical Value: The classical bilinear incidence rate model $\beta IS$ can only describe idealized and simple transmission processes, failing to capture complex dynamical behaviors in the real world (such as sudden outbreaks, persistent epidemics, and periodic oscillations).

Limitations of Existing Methods

1. Restrictions of Simple Models: Classical bilinear incidence rate models typically have at most one endemic equilibrium point, unable to represent complex phenomena such as bistability or periodicity.
2. Parameter-Specific Research: Previous studies on the incidence rate $f(I)S = \beta I(1+\upsilon I^{k-1})S$ have focused mainly on specific values of $k$ (such as $k=1$ or $k=2$).
3. Incomplete Bifurcation Analysis: Lack of systematic research on the complete bifurcation structure of the system for general parameters $k>0$.

Research Motivation

By introducing saturation and nonlinear characteristics, nonlinear incidence rates can capture rich dynamical behaviors arising from complex factors such as psychological behavior, resource constraints, and superspreading events. This paper aims to conduct comprehensive bifurcation analysis for the general parameter case $k>0$, revealing more abundant dynamical phenomena than previous studies.

Core Contributions

1. Systematic Equilibrium Point Analysis: For the general parameter case $k>0$, systematically analyzes the existence, types, and stability of equilibrium points, discovering that the system may have at most 3 positive equilibrium points.
2. High-Codimension Bifurcation Analysis:
   • Proves that the system can undergo codimension-3 Bogdanov-Takens bifurcation when $1<k\leq 2$
   • Proves that the system can undergo codimension-4 Bogdanov-Takens bifurcation when $k>2$
   • Analyzes nilpotent focus bifurcations (codimension 3 and 4)
3. Refined Hopf Bifurcation Analysis:
   • Computes the first four focal values
   • Proves that the system can generate at most 4 limit cycles
   • Determines the number and stability of limit cycles in different parameter regions
4. Complete Bifurcation Diagram: Constructs detailed bifurcation diagrams showing the coexistence and transitions of multiple bifurcation phenomena including saddle-node bifurcation, Hopf bifurcation, and homoclinic bifurcation.
5. Numerical Verification: Provides extensive numerical simulation results that intuitively demonstrate various dynamical behaviors predicted by theory.

Methodology Details

Problem Formulation

Study the SIRS epidemic model with nonlinear incidence rate:

where:

• $S(t), I(t), R(t)$: Numbers of susceptible, infected, and recovered individuals
• $\beta>0$: Transmission coefficient
• $\upsilon>0, k>0$: Nonlinear incidence rate parameters
• $b>0$: Population recruitment rate
• $d>0$: Natural death rate
• $\mu>0$: Recovery rate
• $\delta\geq 0$: Rate of immunity loss

Model Reduction

Through total population analysis $N(t)=S(t)+I(t)+R(t)$, obtain $\lim_{t\to\infty}N(t)=\frac{b}{d}\equiv\Lambda$.

The reduced two-dimensional system:

Through variable transformation $I=\frac{d+\delta}{\beta}x$, $R=\frac{d+\delta}{\beta}y$, $t=\frac{1}{d+\delta}\tau$, obtain the normalized system:

where $p=(d+\delta)^{k-1}\upsilon/\beta^{k-1}$, $\Lambda_0=\beta\Lambda/(d+\delta)$, $\gamma=(d+\mu)/(d+\delta)$, $\eta=\mu/(d+\delta)$.

Basic reproduction number: $R_0 := \Lambda_0/\gamma$

Equilibrium Point Analysis Method

Disease-Free Equilibrium $E_0(0,0)$:

• Stable node when $R_0<1$
• Saddle point when $R_0>1$
• Saddle-node when $R_0=1$

Endemic Equilibrium Points: Determined by analyzing zeros of the function $H(x) = (1+px^{k-1})\left(1-\frac{1+\eta}{\Lambda_0}x\right) - \frac{1}{R_0}$

Depending on different ranges of $k$, the system may have 0-3 positive equilibrium points.

Bifurcation Analysis Techniques

1. Saddle-Node Bifurcation

When $H(x)=0$ and $H'(x)=0$, the equilibrium point degenerates into a saddle-node. Through the center manifold theorem and normal form theory, the paper proves that the system undergoes saddle-node bifurcation under parameter variation.

2. Bogdanov-Takens Bifurcation

When the Jacobian matrix of an equilibrium point has a double zero eigenvalue, the system may undergo Bogdanov-Takens bifurcation. The paper transforms the system into standard normal form through a series of coordinate changes:

and verifies the non-degeneracy condition $\frac{\partial(\eta_1,\eta_2,\eta_3,\eta_4)}{\partial(\lambda_1,\lambda_2,\lambda_3,\lambda_4)}\neq 0$.

3. Hopf Bifurcation Analysis

When $\text{Tr}(J(E))=0$ and $\text{Det}(J(E))>0$, the equilibrium point may be a weak focus. Using the formal series method to compute the first four focal values:

and so on.

By analyzing sign changes of focal values, determine the order of weak focus and the number of limit cycles.

Technical Innovations

1. Unified Parameter Analysis Framework: Unlike previous research on specific values of $k$, this paper establishes a unified analysis framework applicable to general $k>0$.
2. Complete Characterization of High-Codimension Bifurcations:
   • First proves that the model can undergo codimension-4 Bogdanov-Takens bifurcation when $k>2$
   • Discovers codimension-4 nilpotent focus bifurcation
3. Precise Determination of Limit Cycle Numbers: By computing four focal values, proves that the system can have at most 4 limit cycles, which is more precise than previously known results (2 limit cycles).
4. Systematic Classification: Provides complete dynamical classification according to different ranges of $k$ ($0<k<1$, $k=1$, $1<k<2$, $k=2$, $k>2$).

Experimental Setup

Numerical Methods

• Use MATLAB for numerical simulations
• Employ fourth-order Runge-Kutta method for solving differential equations
• Use continuation methods to track bifurcation curves

Parameter Selection

The paper carefully selects parameter combinations for different bifurcation phenomena:

Bogdanov-Takens Bifurcation ($k=2$):

• $\Lambda_0=1.8993$, $p=5.7966$, $\eta=2.3072$
• Observe bifurcation sequences by varying $\gamma$

High-Codimension Bifurcation ($k=3$):

• $\Lambda_0=3.1832$, $p=0.9331$, $\eta=0.6355$
• Demonstrate coexistence of three limit cycles

Hopf Bifurcation:

• Multiple parameter sets showing cases of 1-4 limit cycles

Visualization Methods

1. Phase Portraits: Display trajectory behavior in phase space
2. Bifurcation Diagrams: Identify different bifurcation curves in parameter space
3. Limit Cycles: Intuitively show coexistence of multiple limit cycles

Experimental Results

Main Results

1. Existence and Stability of Equilibrium Points (Section 2)

Theorem 2.1: The properties of disease-free equilibrium $E_0(0,0)$ are completely determined by the basic reproduction number $R_0$

Lemmas 2.2-2.4: Depending on different ranges of $k$, the system may have:

• $k\leq 1$: At most 1 positive equilibrium point
• $1<k\leq 2$: At most 2 positive equilibrium points
• $k>2$: At most 3 positive equilibrium points

2. Degenerate Equilibrium Point Analysis (Section 3)

Theorem 3.1: Existence conditions for saddle-nodes

• When $1<k<2$, $E_2^*$ is a saddle-node
• When $k=2$, $\bar{E}_0^*$ is a saddle-node
• When $k>2$, multiple degenerate equilibrium points may be saddle-nodes

Theorem 3.2: Existence of cusp bifurcation When specific parameter conditions are satisfied, the degenerate equilibrium point is a codimension-2 cusp

Theorem 3.3: High-codimension cusps

• $1<k\leq 2$: Codimension-3 cusp
• $k>2$ and $z=\tilde{z}$: Codimension-4 cusp

Theorem 3.4: Nilpotent focus bifurcation Under specific parameter conditions, the system can undergo codimension-3 or codimension-4 nilpotent focus bifurcation

3. Bifurcation Analysis (Section 4)

Theorem 4.1: Saddle-node bifurcation The system undergoes saddle-node bifurcation near parameter surfaces $SN_1, SN_2, SN_3$

Theorem 4.2: Bogdanov-Takens Bifurcation ($k>2$) The system undergoes codimension-4 Bogdanov-Takens bifurcation when $(\Lambda_0,\gamma,\eta,p)$ approaches $(\tilde{\Lambda}_0,\tilde{\gamma},\tilde{\eta},\check{p})$, containing:

• Codimension-2 cusp bifurcation
• Codimension-3 cusp bifurcation
• Saddle-node bifurcation, Hopf bifurcation, homoclinic bifurcation, etc.

Theorem 4.3: Bogdanov-Takens Bifurcation ($1<k\leq 2$) The system undergoes codimension-3 Bogdanov-Takens bifurcation

4. Hopf Bifurcation (Section 5)

Theorem 5.1: When $R_i=0$ ($i=0,1,2,3$), the equilibrium point is a weak focus of order at most 2

Theorem 5.2: When $l_1=0$, $\frac{3}{2}<k<2$ and $k\neq 1.9839$, the equilibrium point can be a third-order weak focus

Lemma 5.1: When $k=2$, the system can have at most 2 limit cycles

Theorem 5.3: When $k>2$ and $R_i=0$, the system can have at most 3 limit cycles

Theorem 5.4: When $k>2$ and $l_2=0$ or $l_4=0$, the system can have at most 3 limit cycles

Remark 5.1: Numerical results show the system can have 4 limit cycles (see Figure 5.5)

Numerical Verification of Bifurcation Sequences

Figure 4.2 ($k=2$): Demonstrates bifurcation sequence as parameter $\gamma$ decreases:

1. Saddle-node bifurcation ($\gamma=2.6731$)
2. Hopf bifurcation ($\gamma=2.6717$)
3. Homoclinic bifurcation ($\gamma=2.6712$)
4. Double limit cycle ($\gamma=2.6687$)

Figure 4.3 ($k=3$): Demonstrates more complex bifurcation sequence:

1. Saddle-node bifurcation
2. Double limit cycle
3. Triple limit cycle
4. Hopf bifurcation

Numerical Evidence for Limit Cycles

Figure 5.1: Shows that when $\frac{3}{2}<k<2$ and $l_1=0$, the system can have 3 limit cycles

Figure 5.2: Shows 1-2 limit cycles when $k=2$

Figure 5.3: Shows 3 limit cycles when $k>2$

Figure 5.4: Shows 3 limit cycles when $k>2$ and $l_2=0$

Figure 5.5: Shows coexistence of 4 limit cycles (parameters: $k=3$)

Experimental Findings

1. Critical Role of Parameter $k$: The range of $k$ determines the dynamical complexity of the system:
   • $k\leq 1$: Relatively simple dynamics
   • $1<k\leq 2$: Appearance of codimension-3 bifurcations
   • $k>2$: Appearance of codimension-4 bifurcations and more limit cycles
2. Diversity of Limit Cycles: The system can exhibit 1-4 limit cycles, which is rare in epidemic models
3. Richness of Bifurcations: The system exhibits almost all classical bifurcation types: saddle-node, Hopf, homoclinic, Bogdanov-Takens, etc.
4. Impact of Basic Reproduction Number: Changes in $R_0$ lead to stability transitions between disease-free and endemic equilibrium points

Related Work

Classical SIRS Models

• Kermack-McKendrick (1927): Proposed the classical SIR model framework
• Hethcote & van den Driessche (1991): Studied nonlinear incidence rate models

Saturated Incidence Rate Models

• Capasso & Serio (1978): Introduced saturated incidence rate $\frac{\beta IS}{1+\alpha I}$
• Xiao & Ruan (2007): Studied non-monotone incidence rate $\frac{\beta IS}{1+\alpha I^2}$
• Zhou et al. (2007): Studied $\frac{kIS}{1+\beta I+\alpha I^2}$

Nonlinear Incidence Rate Models

• Ruan & Wang (2003): Studied $\frac{\beta I^2 S}{1+\alpha I^2}$, discovering codimension-2 Bogdanov-Takens bifurcation and 2 limit cycles
• Zhang et al. (2022): Studied general saturated incidence rate $\frac{\beta I^p S}{1+\alpha I^q}$
• Cui & Zhao (2024): Studied Bogdanov-Takens bifurcation for general parameters $p, q$

Incidence Rate Studied in This Paper

• Lu et al. (2023): Studied $k=1$ case, discovering codimension-3 Bogdanov-Takens bifurcation
• Jin et al. (2007): Studied $k=2$ case, discovering codimension-2 Bogdanov-Takens bifurcation

Advantages of This Paper

1. Generality: Unified analysis for arbitrary $k>0$, rather than specific values
2. Completeness: Discovers higher-codimension bifurcations (codimension-4)
3. Precision: Proves at most 4 limit cycles, more precise than known results
4. Systematicity: Provides complete bifurcation diagrams and dynamical classification

Conclusions and Discussion

Main Conclusions

1. Equilibrium Point Structure:
   • The system always has a unique disease-free equilibrium $E_0$
   • When $R_0>1$, there may exist 1-3 endemic equilibrium points
   • The number and type of equilibrium points are determined by the range of parameter $k$
2. Bifurcation Phenomena:
   • The system can undergo saddle-node bifurcation, Hopf bifurcation, homoclinic bifurcation
   • Codimension-3 Bogdanov-Takens bifurcation exists ($1<k\leq 2$)
   • Codimension-4 Bogdanov-Takens bifurcation exists ($k>2$)
   • Codimension-3 and codimension-4 nilpotent focus bifurcations exist
3. Limit Cycles:
   • The system can have at most 4 limit cycles
   • The number of limit cycles is determined by the codimension of Hopf bifurcation
   • Different parameter regions can observe 1-4 coexisting limit cycles
4. Epidemiological Significance:
   • Nonlinear incidence rates can produce complex dynamical behaviors
   • The system can exhibit bistability and periodic oscillations
   • Simple control strategies may be ineffective (due to multistability and periodic oscillations)

Limitations

1. Theoretical Completeness:
   • For certain parameter regions (such as $R_i\neq 0$ and $l_1\neq 0$), complete analysis of focal values remains incomplete
   • The existence of 4 limit cycles has only numerical evidence, lacking rigorous theoretical proof
2. Practical Applications:
   • Model assumes total population tends to a constant, may not apply to rapidly changing populations
   • Does not consider spatial heterogeneity and age structure
   • Practical difficulties in parameter estimation not discussed
3. Numerical Verification:
   • Numerical verification of certain high-codimension bifurcations is complex
   • Systematic analysis of parameter sensitivity not provided
4. Model Assumptions:
   • Assumes recovered individuals eventually lose immunity, may not apply to all diseases
   • Ignores disease-induced additional mortality

Future Directions

1. Theoretical Refinement:
   • Complete focal value analysis for all parameter regions
   • Rigorously prove the existence of 4 limit cycles
   • Study global dynamical behavior, such as global stability and global bifurcations
2. Model Extensions:
   • Consider spatial diffusion effects (reaction-diffusion systems)
   • Introduce time delays
   • Consider stochastic perturbations (stochastic differential equations)
   • Incorporate age structure or heterogeneity
3. Application Research:
   • Apply the model to specific diseases (such as COVID-19, influenza, etc.)
   • Study the effectiveness of control strategies (vaccination, quarantine, etc.)
   • Parameter estimation and model validation
4. Computational Methods:
   • Develop more efficient bifurcation analysis algorithms
   • Construct complete bifurcation diagrams in parameter space
   • Study chaos and complex dynamics

In-Depth Evaluation

Strengths

1. Theoretical Depth:
   • Employs advanced dynamical systems theory (center manifold theorem, normal form theory, bifurcation theory)
   • Computes the first four focal values, which is technically very challenging
   • Discovers codimension-4 bifurcations, extremely rare in epidemic models
2. Systematicity:
   • Unified analysis for general parameter $k>0$
   • Provides complete equilibrium point classification and stability analysis
   • Constructs detailed bifurcation diagrams
3. Novelty:
   • First proves the model can have 4 limit cycles
   • First discovers codimension-4 Bogdanov-Takens bifurcation
   • Discovers coexistence of multiple high-codimension bifurcations
4. Numerical Verification:
   • Provides numerous high-quality phase portraits and bifurcation diagrams
   • Numerical results highly consistent with theoretical predictions
   • Clear visualization, easy to understand
5. Writing Quality:
   • Clear logic, reasonable structure
   • Detailed mathematical derivations (though some long expressions are omitted)
   • Rich figures and illustrations with strong explanatory power

Weaknesses

1. Theoretical Gaps:
   • Focal value analysis incomplete for certain parameter regions (such as Section 5.1.2)
   • Rigorous existence proof for 4 limit cycles missing
   • Insufficient global dynamics analysis
2. Practical Applications:
   • Lacks comparison with real epidemic data
   • Does not discuss biological significance and practical measurability of parameters
   • Missing application research on control strategies
3. Technical Details:
   • Many intermediate calculation results omitted ("omitted for brevity")
   • Some symbol definitions not sufficiently clear
   • Insufficient details on specific numerical method implementations
4. Readability:
   • Overly technical for non-expert readers in certain sections
   • Lacks intuitive explanation of biological significance
   • Some theorem conditions excessively complex
5. Experimental Design:
   • Lacks parameter sensitivity analysis
   • No reproducibility materials such as code or data
   • Parameter selection rationale for some numerical results unclear

Impact

1. Academic Contribution:
   • Provides important supplement to bifurcation theory of epidemic models
   • Discovered high-codimension bifurcations and multiple limit cycle phenomena have theoretical value
   • Provides new analytical framework for related research
2. Practical Value:
   • Reveals complex epidemic dynamics that nonlinear incidence rates may produce
   • Provides theoretical support for public health decision-making
   • Indicates that simple control strategies may be ineffective (due to multistability and periodic oscillations)
3. Reproducibility:
   • Detailed theoretical derivations can be verified
   • Clear numerical simulation parameters
   • But lacks code and detailed numerical method descriptions
4. Citation Value:
   • Important reference for researchers studying nonlinear epidemic models
   • Methodological value for mathematicians studying high-codimension bifurcations
   • Likely to become important literature in the field

Applicable Scenarios

1. Theoretical Research:
   • Applications of dynamical systems theory
   • Development of bifurcation theory
   • Mathematical analysis of epidemic models
2. Epidemiological Applications:
   • Diseases with complex transmission mechanisms (such as superspreading events)
   • Diseases requiring consideration of psychological factors and behavioral changes
   • Diseases exhibiting periodic outbreaks
3. Policy Making:
   • Understanding complex disease transmission dynamics
   • Evaluating potential effects of intervention measures
   • Identifying possible transitions and critical points
4. Educational Use:
   • Case studies for dynamical systems courses
   • Application examples of bifurcation theory
   • Teaching materials for mathematical modeling courses

Selected References

This paper cites 37 important references, mainly including:

Classical Epidemic Models:

• 28 Kermack & McKendrick (1927): Classical SIR model
• 18 Levin et al. (1989): Applied mathematical ecology

Nonlinear Incidence Rates:

• 14 Capasso & Serio (1978): Saturated incidence rate
• 15 Xiao & Ruan (2007): Non-monotone incidence rate
• 9 Ruan & Wang (2003): Bifurcations of nonlinear incidence rates

Bifurcation Theory:

• 32 Zhang et al. (1992): Qualitative theory of differential equations
• 33 Khibnik et al. (1998): Global study of cubic Liénard equations
• 34 Dumortier et al. (2001): Unfolding of codimension-4 nilpotent saddles

Related Work:

• 30 Lu et al. (2023): Study of $k=1$ case
• 31 Jin et al. (2007): Study of $k=2$ case
• 20-22 Zhang, Cui, Hu, et al.: Research on general saturated incidence rates

Overall Evaluation: This is a high-quality dynamical systems theory paper that achieves important progress in bifurcation analysis of epidemic models. Theoretical depth and systematicity are its greatest strengths, though there is room for improvement in practical applications and reproducibility. For scholars researching nonlinear dynamical systems and mathematical epidemic models, this is a paper worthy of in-depth study.